Guide · Beginner
History of AI Search: From Featured Snippets to AI Mode
AI search grew out of a decade of steps toward answering questions directly: the Knowledge Graph in 2012, then featured snippets, then the chat assistants and Google's Search Generative Experience in 2022–23, followed by AI Overviews in 2024 and AI Mode in 2025. Each step reduced the need to click and raised the value of being the source an answer draws on.
Why does the history matter?
Many marketers treat AI search as a sudden break. In practice, search engines have been moving from "ten blue links" toward direct answers for more than a decade. Knowing the path helps you see what still works (clear, trustworthy, well-structured information) and what's new (answers that combine many sources and name brands directly).
Timeline
| Date | Milestone | What changed |
|---|---|---|
| May 2012 | Google introduces the Knowledge Graph (Google) | Search starts understanding entities, "things, not strings" |
| Mid-2010s | Featured snippets appear widely (Google) | A quoted passage from one page appears above the results |
| Late 2022 | ChatGPT launches; Perplexity launches its answer engine | Conversational answers become mainstream |
| Early 2023 | Microsoft launches the new Bing with chat | Search plus a language model, with citations |
| 10 May 2023 | Google announces SGE in Search Labs (Google) | Generative answers tested inside Google Search |
| Nov 2023 | GEO research paper posted (arXiv) | The term "generative engine optimization" is coined |
| 14 May 2024 | AI Overviews launch for US users (Google) | Generative answers move from test to default for many queries |
| Late 2024 | ChatGPT adds web search | A major chat assistant cites live web sources |
| 5 Mar 2025 | Google introduces AI Mode in Labs (Google) | A full conversational search mode |
| 20 May 2025 | AI Mode rolls out in the US (Google) | Conversational search becomes widely available |
What did the Knowledge Graph change?
Google described the Knowledge Graph as a way to understand real-world entities such as people, places and things, and the relationships between them (Google). For brands, being a clear, well-described entity became important. The same principle underpins entity SEO for AI today; see Entity SEO for AI.
What did featured snippets change?
Featured snippets reverse the usual format: the descriptive snippet comes first, then the link (Google). They taught marketers to write concise, answer-first passages. That habit carries straight over to AI answers, which also prefer clear, self-contained passages.
What did chat assistants change?
Chat assistants let users ask long, specific questions and follow-ups. Instead of choosing between links, people got a written answer, often naming specific brands. Early versions relied mostly on training data; later versions added live web search and citations. That shift made crawler access and current, accurate pages matter for AI visibility.
What did SGE and AI Overviews change?
SGE brought generative answers into the main Google results page as an experiment (Google). AI Overviews made them a standard feature for many queries in the US in May 2024 (Google). Google says AI Overviews and AI Mode use its search systems and show links to supporting pages, and that no special optimisation beyond normal SEO best practices is required to appear (Google).
What did AI Mode change?
AI Mode offers a full conversational experience inside Google Search, using techniques such as query fan-out to explore subtopics (Google). It moves Google closer to ChatGPT and Perplexity and makes multi-step, follow-up questions part of search.
What has stayed the same?
| Constant | Why it still matters |
|---|---|
| Crawling and indexing | Pages that can't be found can't be used |
| Clear, accurate content | Every answer format favours it |
| Entities and reputation | Engines need to know who you are and whether to trust you |
| Helpful, people-first content | Google's guidance applies to AI features too (Google) |
What has changed?
- Fewer clicks for simple questions, because the answer is on the page
- Brands named directly in recommendations
- Many sources blended into a single answer
- Answers that vary from run to run
- Measurement that has to go beyond rank tracking
A quick example
Illustrative. For the question "how long to boil an egg", a 2010 searcher clicked a recipe site. A 2016 searcher read a featured snippet quoting one site. A 2025 searcher reads an AI Overview combining several sources with links. The publisher that's useful in all three eras is the one with a clear, accurate, well-structured answer.
Checklist
- [ ] Your brand is a clear entity across the web
- [ ] Key pages have concise, answer-first passages
- [ ] Crawlers for AI search are allowed
- [ ] Visibility is measured across engines, not just Google rankings
Related
Sources
- Google, Introducing the Knowledge Graph: things, not strings, 2012
- Google Search Central, Featured snippets and your website
- Google, Search Generative Experience announcement, 2023
- Google, Google I/O 2024: New generative AI experiences in Search, 2024
- Google, Expanding AI Overviews and introducing AI Mode, 2025
- Google, AI Mode in Google Search: updates from Google I/O 2025
- Google Search Central, AI features and your website
- Google Search Central, Creating helpful, reliable, people-first content
- Aggarwal et al., GEO: Generative Engine Optimization
Questions
When did Google launch AI Overviews?
Google announced AI Overviews for everyone in the US at Google I/O on 14 May 2024, after testing them as the Search Generative Experience (SGE) from May 2023.
What came before AI answers?
Direct answers such as the Knowledge Graph (2012) and featured snippets already showed information on the results page without a click.
When did AI Mode launch?
Google introduced AI Mode as a Labs experiment on 5 March 2025 and announced a US rollout on 20 May 2025.